Distribution network grounding fault information research and judgment method and system, computer equipment and medium

Abstract

The invention relates to the technical field of power system analysis, in particular to a distribution network grounding fault information research and judgment method and system, computer equipment and a medium. The method comprises the following steps: acquiring multi-source fault data in a distribution network system and preprocessing the multi-source fault data to obtain a standardized fault data set; extracting fault feature information from the standardized fault data set through a time sequence pattern recognition algorithm; based on the fault feature information, performing clustering analysis based on a topological relation on the fault feature information through a multi-source data fusion research and judgment algorithm, and generating a fault section positioning result; and generating distribution network grounding fault information according to the topological relation identifier and the electrical parameter characteristics in the fault section positioning result. By means of the mode, the technical problem that an existing ground fault research and judgment technology is insufficient in positioning precision in a power distribution network with the high reliability requirement is solved, and the automation and intelligence level of fault research and judgment and the accuracy and usability of a research and judgment conclusion are improved.

Description

Technical Field

[0001] This invention relates to the field of power system analysis technology, and in particular to a method, system, computer equipment, and medium for judging grounding fault information in distribution networks. Background Technology

[0002] In power system operation, single-phase grounding faults are one of the most common fault types in distribution networks, accounting for over 70% of the total system fault rate. Grounding faults not only cause overvoltage surges, leading to insulation breakdown and equipment burnout, but also allow the fault current to continuously flow into the ground, affecting the operating logic of protection devices and potentially inducing cascading trips, causing large-scale power outages, and even chain reactions such as fires and explosions. In particular, prolonged grounding faults severely impact system lifespan and power supply reliability. For example, in 2024, a power supply company in a certain region experienced 40 grounding faults, with an average fault duration of 1.5 hours. This highlights significant problems in the grounding fault assessment process: excessively long assessment times and inaccurate fault location lead to prolonged fault patrol and search times, severely impacting fault handling efficiency.

[0003] Currently, the assessment of grounding faults in distribution networks mainly relies on alarm information transmitted from substation line selection devices and feeder terminals, combined with line topology diagrams for manual or simple automated analysis. Existing technology (application publication number CN119291395A) discloses a method for locating single-phase grounding fault sections in distribution networks based on multi-source data fusion. This method acquires data from the distribution cloud master station and dispatch telemetry data, and uses a combination of grounding fault alarm information, zero-sequence current telemetry values, and grounding fault waveform recordings for comprehensive analysis, aiming to improve the reliability and accuracy of fault location. However, the existing methods still have limitations: First, their judgment strategies are not adaptable enough to different neutral grounding methods (such as systems grounded via arc suppression coils or ungrounded systems), failing to fully consider scenarios of information transmission failure or data loss, thus limiting the accuracy of judgment in practical applications; second, the method has weak perception of global fault characteristics in multi-feeder conditions under complex distribution network topologies, making it difficult to effectively cope with voltage anomalies caused by reverse power flow or intermittent fluctuations; finally, existing technologies generally rely on offline calculations or fixed thresholds, lacking a real-time dynamic adjustment mechanism, which easily leads to judgment delays or misjudgments. Therefore, existing ground fault judgment technologies suffer from insufficient positioning accuracy in distribution networks with high reliability requirements, failing to meet the actual needs of rapid fault isolation and power restoration. Summary of the Invention

[0004] To address the aforementioned shortcomings or deficiencies, this invention provides a method, system, computer equipment, and medium for analyzing and judging grounding fault information in distribution networks. This invention can solve the technical problem of insufficient positioning accuracy in existing grounding fault analysis technologies for distribution networks with high reliability requirements.

[0005] This invention provides a method for judging distribution network grounding fault information, including: Acquire multi-source fault data from the distribution network system and preprocess the multi-source fault data to obtain a standardized fault dataset.

[0006] The time-series pattern recognition algorithm extracts fault feature information from the standardized fault dataset. The time-series pattern recognition algorithm is configured to extract the first half-wave features and measure the waveform similarity of the zero-sequence current waveform in the standardized fault dataset.

[0007] Based on fault feature information, a multi-source data fusion judgment algorithm is used to perform topological clustering analysis on the fault feature information to generate fault segment location results. The multi-source data fusion judgment algorithm is configured to use waveform correlation based on time derivative Euclidean distance to calculate the output segment identifier.

[0008] Distribution network grounding fault information is generated based on the topological relationship identifiers and electrical parameter characteristics in the fault section location results.

[0009] According to a second aspect, the present invention provides a distribution network grounding fault information assessment system, comprising: The fault dataset construction module is used to acquire multi-source fault data in the distribution network system and preprocess the multi-source fault data to obtain a standardized fault dataset.

[0010] The fault feature information extraction module is used to extract fault feature information from the standardized fault dataset through a time-series pattern recognition algorithm. The time-series pattern recognition algorithm is configured to extract the first half-wave features and measure the waveform similarity of the zero-sequence current waveform in the standardized fault dataset.

[0011] The fault segment location module is used to perform topological clustering analysis on the fault feature information based on the fault feature information through a multi-source data fusion and judgment algorithm to generate fault segment location results. The multi-source data fusion and judgment algorithm is configured to use waveform correlation based on time derivative Euclidean distance to calculate and output segment identifiers.

[0012] The grounding fault information generation module is used to generate distribution network grounding fault information based on the topological relationship identifier and electrical parameter characteristics in the fault section location results.

[0013] According to a third aspect, the present invention provides a computer device comprising: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform any of the distribution network grounding fault information judgment methods in the embodiments of the present invention.

[0014] According to another aspect of the present invention, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute any of the distribution network grounding fault information assessment methods in the embodiments of the present invention.

[0015] The present invention provides a method for analyzing distribution network grounding fault information. This method is achieved through four core steps: multi-source data acquisition and standardization, intelligent extraction of time-series features, multi-source information fusion analysis, and structured information generation. Specifically, it acquires and preprocesses multi-source fault data from the distribution network system to integrate heterogeneous information such as distribution terminal grounding alarms, zero-sequence current telemetry values, and fault waveform recordings, transforming it into a standardized fault dataset with a unified format. This solves the problem of weak analytical foundation caused by isolated data sources and inconsistent formats in traditional manual analysis. A time-series pattern recognition algorithm is used to extract fault feature information from the standardized dataset. This algorithm is configured to extract the first half-wave features and measure waveform similarity of the zero-sequence current waveform, accurately capturing key electrical transient features in the early stages of a fault from massive time-series data. This solves the problem of low fault identification caused by insufficient utilization of waveform time-series information and coarse feature extraction in traditional methods. Based on the... The acquired fault feature information is used to perform topological clustering analysis through a multi-source data fusion judgment algorithm. This algorithm is configured to use waveform correlation calculation based on time derivative Euclidean distance to output segment identifiers, which are used to fuse electrical features and power grid topology information to achieve accurate calculation and location of faults. This solves the problems of low location efficiency and accuracy caused by reliance on manual experience inference and lack of multi-source information correlation analysis. Based on the topological relationship identifiers and electrical parameter features in the fault segment location results, distribution network grounding fault information is generated to transform the judgment conclusions into a structured output containing clear fault locations, boundaries, and electrical features. This solves the drawbacks of traditional methods, such as vague conclusions and difficulties in subsequent automated processing.

[0016] In this technical solution, the present invention addresses the problems of insufficient utilization and low analysis efficiency of multi-source fault data as described in the background technology. By constructing a standardized multi-source fault dataset, it provides a unified and high-quality analytical foundation for subsequent algorithms, achieving effective integration and management of data from multiple systems such as power distribution terminals and waveform recording devices. Regarding the issue of fault feature extraction relying on manual methods and difficulty in ensuring accuracy and consistency, the invention automatically performs first-half-wave feature extraction and waveform similarity measurement using a time-series pattern recognition algorithm. This transforms expert experience into a repeatable and quantifiable calculation process, improving the objectivity and reproducibility of feature extraction. Addressing the problem of disconnect between electrical and topological quantity analysis and low reliability of location results during fault segment location, the invention employs a multi-source data fusion analysis algorithm. This combines waveform-correlation-based electrical quantity analysis with switch-association-based topological clustering analysis, achieving information complementarity and cross-validation, significantly improving the accuracy and reliability of location analysis. Finally, addressing the issue of non-standardized analysis results output that cannot directly support automated operation and maintenance decisions, the invention automatically encapsulates the location results and electrical parameters into structured fault information, generating a machine-readable and clearly defined standardized analysis report. Therefore, the technical solution of the present invention solves the technical problem of insufficient positioning accuracy of existing ground fault judgment technology in distribution networks with high reliability requirements, and improves the automation and intelligence level of fault judgment as well as the accuracy and usability of judgment conclusions. Attached Figure Description

[0017] Figure 1 This is a flowchart of a distribution network grounding fault information assessment method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the overall architecture of a power outage and power restoration information monitoring and reporting system for a distribution network area, according to another embodiment of the present invention. Figure 3 This diagram illustrates the specific implementation flowchart of a security data acquisition agent performing a real-time fault data acquisition task according to another embodiment of the present invention. Figure 4 This is a schematic diagram of the structure of a distribution network grounding fault information analysis system according to an embodiment of the present invention; Figure 5 This is a block diagram of a computer device for implementing embodiments of the present invention. Detailed Implementation

[0018] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention, including various details to aid understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0019] During the development of this invention, researchers conducted numerous experiments and data analysis, discovering an intrinsic correlation between the characteristics of the first half-wave of the zero-sequence current waveform in the early stages of a fault and the fault location: the first half-wave waveforms of the zero-sequence current collected from monitoring points on both sides of the fault point show significant differences in the shape of their time derivatives, and this difference is strongly correlated with the electrical distance from the fault point to the monitoring point. Based on this relationship, this invention innovatively proposes this technical solution, which uses a time-sequence pattern recognition algorithm to quantitatively extract waveform features, performs topological clustering analysis through a multi-source data fusion and judgment algorithm, and combines this with power grid topology information to achieve accurate and automatic location of the fault location, embodying the core concept of "feature-driven, topology-constrained, and intelligent judgment".

[0020] Specifically, through comparative experiments, the invention team discovered that traditional methods relying on human experience and combining single electrical quantity thresholds for judgment suffer from technical defects such as low efficiency, poor consistency, and difficulty in handling complex faults. Their judgment results are highly dependent on personal experience, failing to effectively integrate waveform timing details and network topology relationships, resulting in limited positioning accuracy and lengthy processing times. In contrast, the distribution network grounding fault information judgment method proposed in this invention improves the automation and intelligence of the judgment process. By standardizing multi-source fault data and extracting key waveform features using a timing pattern recognition algorithm, objective and quantitative characterization of fault features can be achieved. By fusing electrical features and network structure information through topological clustering analysis, the accuracy and reliability of fault segment location results can be ensured. Furthermore, by generating structured distribution network grounding fault information, standardized output and efficient transfer of judgment conclusions can be achieved, directly supporting subsequent automated operation and maintenance decisions.

[0021] Therefore, this invention provides a method for judging distribution network grounding fault information based on the first aspect. This method can be applied to a distribution network fault intelligent judgment and auxiliary decision-making system (hereinafter referred to as the "system"). The system can run on the main station server of the distribution network dispatch center, cloud computing platform, or edge computing device through independent software applications or integrated service components to complete the fully automated judgment process from multi-source fault data access, intelligent feature extraction, fault section location to structured report generation. Specifically, this system can be deployed in various hardware environments, including but not limited to: physical servers or server clusters deployed in municipal-level dispatch centers, virtualized computing resources relying on power private clouds or hybrid clouds, and edge computing gateways installed in substations or distribution automation terminals. This flexible deployment architecture enables the system to meet both the performance requirements of the dispatch center for centralized processing of the entire network data and large-scale parallel computing, and the lightweight judgment requirements of the field or regional edge side under low latency, high real-time performance, and limited resource constraints.

[0022] like Figure 1 As shown, the method may include: Step S110: Obtain multi-source fault data from the distribution network system and preprocess the multi-source fault data to obtain a standardized fault dataset.

[0023] Multi-source fault data refers to heterogeneous data from multiple sources, such as distribution terminal grounding alarm signals, bus grounding information, low-current grounding line selection information, arc suppression coil information, and bus voltage information, obtained from the distribution network IV zone system. Standardized fault datasets refer to unified format datasets formed after data cleaning, format conversion, and outlier filtering, with a data error rate not exceeding 1%. The distribution network IV zone system is a distribution automation master station system built according to the power system safety zoning principle, used for managing large information zones.

[0024] Specifically, the system can establish a secure communication tunnel with the distribution network IV zone system through a secure data acquisition agent (a software module that simulates manual operation). It obtains a data access token through a credential authentication mechanism and uses a human-computer interaction simulation engine to parse a preset graphical interface operation script to drive the data acquisition process. Subsequently, it performs hash verification (a data integrity verification algorithm) on the acquired raw data stream and uses the Transport Layer Security (TLS) protocol to build an end-to-end data security pipeline to ensure transmission reliability. Finally, it inputs the encrypted data into the data preprocessing pipeline to perform data pattern mapping and transformation (such as converting Excel format to CSV format) and outlier filtering based on statistical distribution (such as removing data points that deviate from the mean by more than three standard deviations).

[0025] For example, the system downloads 10 ground fault data from the distribution network IV area system, with a daily data volume of about 5 megabytes (MB). The data is integrated using the Python Pandas library (an open-source data analysis and processing tool library based on the Python programming language). After cleaning, the average data error rate is 0.02%, and the generated standardized fault dataset file is about 2MB in size, which can be directly used for subsequent analysis.

[0026] Step S120: Extract fault feature information from the standardized fault dataset using a time-series pattern recognition algorithm.

[0027] The time-series pattern recognition algorithm is configured to extract the first half-wave feature and measure waveform similarity of the zero-sequence current waveform in the standardized fault dataset. The first half-wave feature extraction refers to extracting the first half-wave of the zero-sequence current waveform at the initial stage of the fault occurrence (the time range is usually 0 to 20 milliseconds). Waveform data; waveform similarity measurement refers to quantifying the differences in waveform morphology by calculating the Euclidean distance (a mathematical method for measuring the distance between vectors) of the waveform time derivative. The smaller the distance value, the higher the similarity.

[0028] Specifically, the system can extract the zero-sequence current first half-wave waveform from ground fault recording data through first half-wave truncation processing. Then, it uses a nonlinear least squares optimization algorithm (a parameter fitting method) to perform nonlinear fitting on the discrete points of the waveform to obtain a zero-sequence current fitting function (such as a quadratic polynomial function). Next, it calculates the waveform time derivative parameter (first derivative) based on the fitting function and performs Euclidean distance calculation based on the derivative sequence, as shown in the formula: ;in, and These represent the two waveforms at different time points. The derivative value, The number of sampling points (e.g., 64 points) is used to obtain the waveform similarity metric.

[0029] For example, the system processes a 64×64 dimension zero-sequence current matrix (64 subcarriers and 64 time sampling points). After the first half-wave is truncated, the waveform length is 32 points, and the calculation of the fitting function takes about 5 milliseconds. The Euclidean distance calculation results are used to quantify the similarity of different switching waveforms. The similarity metric ranges from 0 to 100 (dimensionless), and a value below 10 indicates high similarity.

[0030] Step S130: Based on the fault feature information, perform topological clustering analysis on the fault feature information through a multi-source data fusion and judgment algorithm to generate fault segment location results.

[0031] Among them, the multi-source data fusion judgment algorithm is configured to use waveform correlation calculation based on Euclidean distance of time derivative to output segment identifiers. Waveform correlation calculation refers to converting waveform similarity measurement values ​​into correlation coefficients (such as Pearson correlation coefficient) to quantify the waveform correlation strength between feeder outgoing switches. Segment identifiers are symbols used to uniquely identify fault segments (such as switch number "S001"). Cluster analysis based on topology refers to grouping switch nodes according to the power grid topology diagram (a graph structure that describes the connection relationship of switches) to identify fault boundaries.

[0032] Specifically, the system can construct a multi-dimensional feature vector (with 5 dimensions, including waveform similarity metrics, mean time derivative, peak value, etc.) for the feeder outgoing switches, and calculate the waveform correlation coefficient between each switch using a multi-dimensional feature distance statistical algorithm (the formula is: ,in, and For feature vector elements, and (The mean is used). Then, feeder-level clustering analysis (such as K-means clustering algorithm, K=3) is used to determine the feeder with the lowest waveform correlation coefficient as the faulty line. Next, based on the topology of the faulty line, the waveform correlation coefficient between adjacent switch nodes is calculated by segment-level graph analysis algorithm, and the adjacent switch pairs with the lowest correlation coefficient are identified by minimum similarity node pair identification algorithm (an optimization method based on graph traversal), generating the fault segment location result.

[0033] For example, the system processes data from 10 feeders, and the cluster analysis takes approximately 2 seconds. The faulty line "L005" was identified. Its waveform correlation coefficient with the adjacent switch was 0.15 (below the threshold of 0.3). The faulty section was located between switches "S012" and "S013". The section identifier output was "FZone_012-013".

[0034] Step S140: Generate distribution network grounding fault information based on the topological relationship identifier and electrical parameter characteristics in the fault section location results.

[0035] Among them, the topology identifier refers to the number of the fault section boundary switch (such as "S012-S013"); the electrical parameter characteristics include the effective value of the zero-sequence current (unit: amperes). The fault information includes the fault location, electrical parameters, and timestamp, and is used to support automated operation and maintenance decisions.

[0036] Specifically, the system can map switch numbers to standardized identifiers (e.g., "Border_S012") using a boundary node encoding algorithm. A feature extraction engine extracts line identifiers (e.g., "10kV_L005") and fault phase characteristics from the location results. Combined with timestamp information (e.g., "2025-09-14 10:30:00"), an event description generator constructs a standardized fault event description (text format). Finally, the above information is input into the information fusion module, which performs format standardization encapsulation using a data serialization protocol (e.g., JSON serialization) to output a distribution network grounding fault information file. JSON stands for "JavaScript Object Notation." It is language-independent and is a completely text-based format that is easy for humans to read and write, as well as easy for machines to parse and generate.

[0037] For example, the system generates fault information of 50 kilobytes in size. The file contains the following fields: {Line Name: "Renfu E606 Line", Phase: "C", Fault Time: "2025-07-15 08:15:00", Section Identifier: "S012-S013"}. This file can be distributed via SMS or a collaborative office system, with a processing time of less than 30 seconds.

[0038] In other embodiments, such as Figure 2 This demonstrates the overall architecture of a power outage and restoration information monitoring and reporting system for distribution network areas. This system integrates the distribution network grounding fault information assessment method described in this invention as its core analysis and decision-making engine. The architecture clearly outlines the entire process from data source to information output using a tree-like logic. Specifically, the "data refresh module" (corresponding to its sub-module "waiting for re-looping") periodically triggers data acquisition tasks. Its function corresponds to the security data acquisition agent in the following embodiments, responsible for downloading multi-source fault data from the "Distribution Network IV Zone System". The acquired raw data then enters the "data integration module" (implemented in Python), which performs the aforementioned preprocessing and feature extraction processes, constructs a standardized fault dataset, and generates fault feature information. The structured fault information generated (including fault sections, lines, phases, etc.) then flows to the "information editing module" (supporting custom SMS editing) and the "information sending module" (with a Python interface display). These two modules collaboratively implement the human-computer interaction interface functions defined below, completing information visualization, interactive review, and multi-channel distribution. Finally, the "Process Anomaly Handling Module" (whose sub-module is the "Wait and Restart Module") forms the cornerstone of system reliability. Its periodic monitoring and anomaly restart mechanisms are technically identical to the system reliability assurance architecture described below, ensuring the continuous and stable operation of the entire assessment and reporting process. Therefore, Figure 2 The system architecture fully illustrates the composition, collaborative relationship, and application form of each core functional module in specific business scenarios (power outage and power restoration information monitoring and reporting) when the judgment method of this invention is engineered and deployed in an actual production system.

[0039] Therefore, according to the above implementation method, the system achieves its goals through four core steps: multi-source data acquisition and standardization, intelligent extraction of time-series features, multi-source information fusion and analysis, and structured information generation. Specifically, it acquires and preprocesses multi-source fault data from the distribution network system to integrate heterogeneous information such as distribution terminal grounding alarms, zero-sequence current telemetry values, and fault waveform recordings, transforming it into a standardized fault dataset with a unified format. This solves the problem of weak analytical foundation caused by isolated data sources and inconsistent formats in traditional manual analysis. A time-series pattern recognition algorithm is used to extract fault feature information from the standardized dataset. This algorithm is configured to extract the first half-wave features and measure waveform similarity of the zero-sequence current waveform, accurately capturing key electrical transient features in the early stages of a fault from massive time-series data. This solves the problem of low fault identification caused by insufficient utilization of waveform time-series information and coarse feature extraction in traditional methods. Based on the... The acquired fault feature information is used to perform topological clustering analysis through a multi-source data fusion judgment algorithm. This algorithm is configured to use waveform correlation calculation based on time derivative Euclidean distance to output segment identifiers, which are used to fuse electrical features and power grid topology information to achieve accurate calculation and location of faults. This solves the problems of low location efficiency and accuracy caused by reliance on manual experience inference and lack of multi-source information correlation analysis. Based on the topological relationship identifiers and electrical parameter features in the fault segment location results, distribution network grounding fault information is generated to transform the judgment conclusions into a structured output containing clear fault locations, boundaries, and electrical features. This solves the drawbacks of traditional methods, such as vague conclusions and difficulties in subsequent automated processing.

[0040] Specifically, in the technical solution of this embodiment, to address the problems of insufficient utilization and low analysis efficiency of multi-source fault data as described in the background technology, a standardized multi-source fault dataset is constructed, providing a unified and high-quality analysis foundation for subsequent algorithms, and realizing the effective integration and management of data from multiple systems such as power distribution terminals and waveform recording devices. To address the problem of fault feature extraction relying on manual labor and difficulty in ensuring accuracy and consistency, a time-series pattern recognition algorithm is used to automatically perform first-half-wave feature extraction and waveform similarity measurement, transforming expert experience into a repeatable and quantifiable calculation process, thus improving the objectivity and reproducibility of feature extraction. To address the problem of disconnect between electrical quantity and topology quantity analysis and low reliability of location results during fault section location, a multi-source data fusion judgment algorithm is adopted, combining waveform-correlation-based electrical quantity analysis with switch-association-based topology clustering analysis, achieving information complementarity and cross-validation, significantly improving the accuracy and reliability of location judgment. To address the problem of non-standard judgment result output that cannot directly support automated operation and maintenance decisions, the location results and electrical parameters are automatically encapsulated into structured fault information, generating a machine-readable and clearly defined standardized judgment report. Therefore, the technical solution of this embodiment solves the technical problem of insufficient positioning accuracy of existing ground fault judgment technology in distribution networks with high reliability requirements, and improves the automation and intelligence level of fault judgment as well as the accuracy and usability of judgment conclusions.

[0041] In some embodiments, the standardized fault dataset includes ground fault waveform data; extracting fault feature information from the standardized fault dataset using a time-series pattern recognition algorithm includes: The zero-sequence current waveform in the ground fault recording data is processed by first half-wave truncation to obtain the first half-wave waveform.

[0042] The first half-wave truncation process refers to extracting the first complete half-cycle waveform data after the fault initiation time (denoted as t0) from the zero-sequence current time-series data recorded in the fault waveform file. This process aims to focus on the transient current information that best reflects the fault characteristics in the early stage of the fault occurrence.

[0043] Specifically, the system can use the data slicing module to locate faults based on the time stamp of the fault initiation signal. At a specific time, and at a preset fixed length (e.g., half of 20 milliseconds at the power frequency, i.e., 10 ms) or by dynamically detecting zero-crossing points, the system extracts corresponding data segments from the original waveform array. For example, the system identifies a segment of zero-sequence current waveform data with 4096 sampling points and a sampling rate of 10 kHz from the waveform. The data from sampling point 1024 to sampling point 1535 (a total of 512 points, corresponding to 10.24ms) is then taken as the first half-wave waveform.

[0044] The zero-sequence current fitting function is obtained by performing nonlinear fitting calculation on the first half-wave waveform using a nonlinear least squares optimization algorithm.

[0045] Among them, the nonlinear least squares optimization algorithm is a parameter estimation method that minimizes the sum of squared residuals between the model's predicted values ​​and the actual observed values ​​through iterative optimization. It is used to reconstruct a continuous function expression from discrete waveform sampling points. The zero-sequence current fitting function is a mathematical function that can characterize the waveform variation law of the first half-wave with high accuracy, and it is usually in polynomial form.

[0046] Specifically, the system can utilize numerical computation libraries (such as Python's SciPy library, an open-source scientific and technical computing toolkit based on the Python programming language) to call the Levenberg-Marquardt algorithm (an iterative optimization algorithm for solving nonlinear least squares problems). It takes the discrete-time-current pairs of the first half-wave waveform as input and optimizes the coefficients of the fitting function to make the function curve as close as possible to all the original data points. For example, the system performs a second-order polynomial fitting on a first half-wave waveform containing 512 points, obtaining the fitting function as follows: ,in This is a relative time measured in milliseconds (ms). The zero-sequence current value is expressed in amperes (A), and the coefficient of determination of this function relative to the original waveform is ( The value reached 0.98.

[0047] The time derivative parameters of the fitted waveform are calculated based on the zero-sequence current fitting function, and the Euclidean distance is calculated based on the time derivative parameters to obtain the waveform similarity metric.

[0048] The time derivative parameter refers to the sequence of first derivatives obtained by differentiating the fitted function, which characterizes the rate of change of the zero-sequence current with time. Euclidean distance is used to measure the straight-line distance between two derivative sequences in vector space. This distance value is a waveform similarity metric; the smaller the value, the more similar the changes in the two waveforms.

[0049] Specifically, the system can calculate the fitting function using either symbolic differentiation or numerical differentiation methods. derivative function The derivative arrays are obtained by sampling the sequence at the same time points. For two waveforms A and B that need to be compared, their derivative arrays are calculated respectively, and the Euclidean distance is calculated according to the formula: ,in and waveforms and In the The derivative value at each time point This represents the summation over all sampled points. For example, the system calculates the derivatives of the fitting functions for the first half-wave of two switches, obtaining two derivative arrays of length 100 by sampling at 100 uniform time points. The Euclidean distance between them is calculated to be 15.6 (dimensionless), which is the similarity measure of the two waveforms.

[0050] Fault characteristic information is constructed based on the zero-sequence current fitting function, time derivative parameters, and waveform similarity metrics.

[0051] Among them, fault characteristic information is a set of quantitative features used for subsequent fault assessment. The zero-sequence current fitting function describes the overall shape of the waveform, the time derivative parameter characterizes the details of local changes in the waveform, and the waveform similarity metric provides a scalar index of the degree of similarity between waveforms at different monitoring points. Together, these three constitute a comprehensive description of the electrical characteristics of the fault.

[0052] Specifically, the system can encapsulate the coefficient array of the fitted function, the derivative parameter array, and multiple waveform similarity metrics calculated for different switch pairs into a structured feature dictionary or object using data structure encapsulation methods, for subsequent algorithm modules to call. For example, the feature information generated by the system for a fault event includes: (1) Fitting function coefficient vector ; (2) An array consisting of 100 derivative values; (3) A 5x5 similarity measure matrix, matrix elements Representing the Switch No. and No. Similarity metric for the waveform of switch number 1.

[0053] Therefore, according to the above implementation method, the system can automatically and accurately extract quantitative feature information that characterizes the nature of the fault from the ground fault recording data, laying a reliable data foundation for intelligent judgment based on multi-source data fusion.

[0054] In some embodiments, based on fault feature information, a multi-source data fusion and analysis algorithm is used to perform topological clustering analysis on the fault feature information to generate fault segment location results, including: Based on waveform similarity metrics, a multidimensional feature vector of the feeder outgoing switches is constructed, and the waveform correlation coefficient between each feeder outgoing switch is calculated using a multidimensional feature distance statistical algorithm.

[0055] Among them, the multidimensional feature vector refers to a vector-based data structure composed of multiple feature indicators, used to comprehensively characterize the electrical behavior of each feeder outgoing switch. Feature dimensions may include waveform similarity metrics, peak current, waveform steepness, etc.; the waveform correlation coefficient is a scalar value obtained through mathematical calculation that quantifies the similarity of waveforms between two switches, with a value range of [range missing]. The closer the value is to 1, the stronger the positive correlation.

[0056] Specifically, the system can use a feature engineering module to extract five preset feature indicators (such as waveform similarity metric, mean time derivative, effective current value, waveform distortion rate, and fault initiation phase) for each feeder outgoing switch, forming a 5-dimensional feature vector. Subsequently, a multi-dimensional feature distance statistical algorithm (such as Euclidean distance or cosine similarity algorithm) is used to calculate the distance or similarity between the vectors, and the distance values ​​are mapped to a correlation coefficient. The formula can be expressed as: ,in, For Euclidean distance, This is a preset maximum distance normalization factor. For example, when the system processes data from the outgoing switches of 10 feeders, a 5-dimensional feature vector is constructed for each switch. For instance, the feature vector for switch S001 is... (Dimensionless); The Euclidean distance between switches S001 and S002 is calculated to be 8.2, and the waveform correlation coefficient after normalization is 0.67.

[0057] Based on the waveform correlation coefficient, the feeder with the lowest waveform correlation coefficient was identified as the faulty line through feeder-level cluster analysis.

[0058] Among them, feeder-level clustering analysis refers to grouping feeders based on the waveform correlation coefficient matrix of all feeder outgoing switches using an unsupervised machine learning algorithm, and identifying the feeder with the greatest waveform characteristic difference as a faulty line; a faulty line refers to a power feeder that has experienced a ground fault, whose switch waveform has a significantly low correlation with the waveforms of other feeder switches.

[0059] Specifically, the system can use clustering algorithms (such as K-means clustering or hierarchical clustering) to take the average waveform correlation coefficient of each feeder as input features, set the number of clusters to 2 (faulty and non-faulty), iteratively optimize the cluster centers, and finally determine the group with the lowest cluster center value as the faulty line. For example, the system calculates the average waveform correlation coefficients of 10 feeders as follows: Feeder ; Using K-means clustering (K=2) to (Coefficient 0.15) is classified as a faulty line, and the remaining feeders are classified as non-faulty groups.

[0060] Based on the switch topology of the faulty line, the waveform correlation coefficient between adjacent switch nodes in the topology graph is calculated using a segment-level graph analysis algorithm.

[0061] Among them, the segment-level graph analysis algorithm refers to abstracting the power grid topology into a graph model (a mathematical modeling and calculation method for analyzing and processing the connection relationships of power systems), where nodes represent switching devices and edges represent electrical connection relationships. The waveform correlation is calculated by traversing adjacent node pairs in the graph; adjacent switching nodes refer to switch pairs that are directly connected by lines in the topology graph.

[0062] Specifically, the system can store topological relationships through a graph database or adjacency matrix, and use a breadth-first search (BFS) algorithm to traverse all adjacent switch pairs of the faulty line, extracting the waveform correlation coefficient for each switch pair (calculated using the same method as the feeder level). For example, the faulty line... Includes switch nodes The topology is The system calculates adjacent pairs. The waveform correlation coefficient is 0.32. The coefficient is 0.18.

[0063] The algorithm for identifying the least similar node pairs identifies the adjacent switch pairs with the lowest waveform correlation coefficient, generating fault section location results. The fault section location results include the identification information of the adjacent switch pairs.

[0064] Among them, the minimum similarity node pair identification algorithm refers to selecting the switch pair with the smallest coefficient value as the fault section boundary by comparing the waveform correlation coefficients of all adjacent switch pairs; the fault section location result refers to the fault location information output in a standardized format, including the boundary switch number, fault line identification, etc.

[0065] Specifically, the system can use a sorting algorithm (such as quicksort) to sort the waveform correlation coefficients of all adjacent switch pairs in ascending order, select the top-ranked switch pair as the fault section, and extract its switch identifier (e.g., "S0102-S0103"). For example, the system identifies the switch pair with the lowest coefficient among adjacent switch pairs as... With a coefficient of 0.18, the fault location result is as follows: Fault line: “L010”, Section boundary: “S0102−S0103”, Correlation coefficient: 0.18.

[0066] Therefore, according to the above implementation method, the system can accurately locate the grounding fault section of the distribution network through multi-level clustering and graph analysis technology, thereby improving the efficiency and accuracy of judgment.

[0067] In some embodiments, distribution network grounding fault information is generated based on the topological relationship identifier and electrical parameter characteristics in the fault section location results, including: The fault section boundary identifier is generated based on the identification information of adjacent switch pairs using a boundary node coding algorithm.

[0068] The boundary node encoding algorithm refers to a rule-based processing method that maps the original identifiers of adjacent switch pairs to standard format strings, aiming to generate unique and meaningful fault section boundary identifiers. Fault section boundary identifiers are text labels used to clearly indicate the starting and ending switches of the fault in the output results.

[0069] Specifically, the system can use string processing functions to extract the switch numbers of two adjacent switch pairs, concatenate them according to preset format rules, and add a connector in the middle. Common encoding rules include: arranging the switch numbers in topological order, connecting them with specific separators (such as hyphens "-"), and possibly adding a prefix to identify their attributes (such as "Border_"). For example, if the system identifies adjacent switch pairs as switches S012 and S013, the fault section boundary identifier can be generated as "Border_S012-S013" using a boundary node encoding algorithm.

[0070] The feature extraction engine extracts the line identifier and fault phase features from the fault section location results.

[0071] The feature extraction engine is a software module specifically responsible for locating and retrieving specific field values ​​from structured data objects by name or path. The line identifier is a string that uniquely identifies a distribution feeder, typically derived from the naming conventions of the power grid dispatching system (e.g., "10kV_Renfu E606 line"). The fault phase characteristic is a textual symbol indicating which phase or phases a ground fault occurred in (e.g., "phase A", "phases BC").

[0072] Specifically, the system can access the fault location result data structure (such as a Python dictionary or JSON object) and directly read the corresponding values ​​using predefined keys, such as "faulty_line" and "fault_phase". For example, if the fault location result object contains {"faulty_line":"10kV_RenfuE606 line","fault_phase":"C",...}, after the feature extraction engine reads it, it obtains the line identifier as "10kV_RenfuE606 line" and the fault phase characteristic as "C".

[0073] By combining timestamp information, a standardized fault event description is constructed using an event description generator.

[0074] The event description generator is a module that automatically generates grammatically correct and standardized natural language description text based on key input information (line, phase, time, section). A standardized fault event description is a text with a uniform format and complete elements, used to summarize the core information of the fault event. Timestamp information refers to the precise time of fault occurrence or assessment completion, typically in the format "year-month-day hour:minute:second".

[0075] Specifically, the system can use string template formatting technology to replace placeholders (such as {line}, {phase}, {time}, {section}) in a preset sentence template with actual values ​​obtained from the feature extraction engine and time module. For example, the system uses the template: "At {time}, {line} experienced a {phase} phase-to-ground fault, and the suspected fault section is {section}." Combining the timestamp "2025-07-15 08:15:00", line identifier, fault phase characteristics, and fault section boundary identifier, the generated standardized fault event description is: "At 2025-07-15 08:15:00, the 10kV_Renfu E606 line experienced a C-phase-to-ground fault, and the suspected fault section is Border_S012-S013."

[0076] The fault section boundary identifier, line identifier, fault phase characteristics, and standardized fault event description are injected into the information fusion engine. The data serialization protocol is used to execute the standardized encapsulation format and output the distribution network grounding fault information.

[0077] The information fusion engine is a software component responsible for integrating multiple heterogeneous data items into a single, structured data object. A data serialization protocol is a set of rules (such as JSON or XML) that convert in-memory data structures into a standardized byte stream or text format that can be stored or transmitted. Distribution network grounding fault information refers to the final output structured data file or message containing complete fault assessment conclusions. XML (Extensible Markup Language) is a general-purpose markup language specification for encoding documents and data.

[0078] Specifically, the system can create a new data structure (such as a dictionary or an instance of a specific class) through the information fusion engine, using all the information obtained in the preceding steps as attributes or key-value pairs of this structure. Subsequently, a serialization library (such as Python's json library, a standard built-in library for encoding and decoding JSON data) is called to convert this data structure into a string or byte stream of the target format. For example, the information fusion engine generates a Python dictionary: {"boundary_id":"Border_S012-S013","line_id":"10kV_Renfu E606 line","fault_phase":"C","event_description":"At 08:15:00 on 2025-07-15, a C-phase ground fault occurred on the 10kV_Renfu E606 line, with the suspected fault section being Border_S012-S013."}. After encapsulation using the JSON serialization protocol, the output distribution network grounding fault information is a JSON string with a file size of approximately 500 bytes (B).

[0079] Therefore, according to the above implementation method, the system can automatically and in a standardized manner generate distribution network grounding fault information containing key judgment conclusions, which significantly improves the efficiency, accuracy and standardization of information output, and provides a reliable data foundation for subsequent fault handling, information distribution and archiving.

[0080] In some embodiments, distribution network grounding fault information is output through a human-machine interface module, which is configured as follows: A graphical monitoring panel is built using a visualization rendering engine to dynamically render the faulty line topology diagram and electrical parameter waveform sequences.

[0081] The visualization rendering engine refers to a dedicated rendering component developed based on graphics libraries (such as Python's Matplotlib or the Web's ECharts), responsible for converting abstract electrical data and topological relationships into an intuitive graphical interface. A graphical monitoring panel refers to a comprehensive display interface integrating various visualization elements, including but not limited to single-line power grid diagrams, waveform graphs, and data tables. Dynamic rendering refers to the technical process of automatically updating the graphical display content based on real-time data changes.

[0082] Specifically, the system can be implemented using Python or JavaScript code by calling the Application Programming Interface (API) of the visualization library. For the topology diagram, a force-directed graph algorithm is used to automatically calculate the switch node layout; for waveform sequences, a line graph component is used to draw curves showing the changes of parameters such as zero-sequence current and voltage over time in real time. The panel supports interactive operations such as zooming and panning. For example, the system uses the ECharts library (an open-source data visualization chart library based on the JavaScript programming language) to build the monitoring panel, displaying 10 kV ( The single-line topology diagram of the distribution network uses different colors to highlight the faulty line "Renfu E606 line" and its upstream and downstream switches; the right side dynamically displays the zero-sequence current waveform collected by the outgoing switch of the line. The waveform data is obtained from the server and refreshed every 5 seconds. The horizontal axis is time (unit: milliseconds) and the vertical axis is the current value (unit: amperes).

[0083] It integrates an interactive editing workflow, supports rich text markup language editing, and structured data verification protocols.

[0084] Interactive editing workflows refer to a series of steps that allow users to view, modify, and confirm automatically generated fault assessment results through a graphical interface. Rich text markup languages ​​(such as HTML) are computer languages ​​that support text formatting. Structured data verification protocols refer to procedural rules that ensure users have reviewed all critical data items before verification and that the verification operation is irreversible. HTML (HyperText Markup Language) is a standard markup language used to create web pages and web applications, not a programming language.

[0085] Specifically, the system can embed a rich text editor (such as TinyMCE, a rich text editor component library based on JavaScript and HTML) into the monitoring panel, providing editing functions such as font, color, and annotation. After the user completes the editing, the system pops up a structured confirmation dialog box containing all key fields, requiring the user to check each item and click the confirmation button. The confirmation operation records the operator's identity, timestamp, and generates a digital signature. For example, the dispatcher modifies the automatically generated description of "suspected fault section" to "confirmed fault section" in the rich text editor and adds a red annotation "confirmed after waveform verification." After the editing is completed, the system pops up a confirmation dialog box, listing key information such as "faulty line: 10kV Renfu E606 line" and "faulty phase: C phase." After the dispatcher confirms that everything is correct, they click the "send" button to complete the process.

[0086] Deploy a multi-channel message distribution engine to encapsulate standardized fault information into a cross-platform message format, adapting to the protocol conversion of SMS gateways and collaborative office systems.

[0087] Among them, a multi-channel message distribution engine refers to a software module that can convert the same information content into multiple formats and send it through different channels according to the requirements of the target platform. Cross-platform message formats refer to standardized data formats that can be recognized by heterogeneous systems (such as SMS platforms and OA systems), such as JSON or XML. Protocol conversion refers to the technical process of encapsulating data from one communication protocol to another.

[0088] Specifically, the system receives standardized fault information through message queues (such as RabbitMQ, an open-source message broker software based on the AMQP standard), and then calls different format converters according to the preset sending target. For SMS channels, the information is compressed into Protocol Data Unit (PDU) format that conforms to the requirements of the operator's SMS gateway (such as China Mobile's CMPP protocol); for collaborative office systems, it is encapsulated into JSON messages required by specific platforms (such as DingTalk robots) and sent via HTTPS (Hypertext Transfer Protocol Secure). For example, the engine converts the fault information into two versions: one is a 140-byte PDU format SMS prepared for the Huawei Cloud SMS gateway, with the content "[Fault Alarm] 10kV Renfu E606 line C phase grounded, section S012-S013"; the other is a JSON message prepared for the DingTalk group robot, containing a detailed fault description in Markdown format, and sent to DingTalk's Webhook (a lightweight integration mechanism based on the HTTP protocol to achieve automated event notification between applications) address via an HTTPS POST request.

[0089] Deploy an asynchronous audit workflow engine to integrate multi-level verification nodes into the message distribution pipeline and execute authentication protocols and digital signature algorithms.

[0090] Among these, the asynchronous audit workflow engine refers to a task scheduling component that supports non-blocking, multi-step audit processes. The message distribution pipeline refers to the series of processing steps a message goes through from generation to delivery. Multi-level verification nodes refer to multiple checkpoints set up in the pipeline. An authentication protocol refers to a system of rules used to verify the identity of an operator. A digital signature algorithm refers to a cryptographic method used to ensure information integrity and non-repudiation.

[0091] Specifically, the system defines the review process through a workflow engine (such as Activiti), for example, "generate message → junior scheduler review → senior scheduler approval → send". At each verification node, the engine calls the unified identity authentication service to verify the validity of the operator token and uses the scheduler's private key to execute a digital signature algorithm (such as RSA with SHA-256) to generate a signature digest for the message payload. For example, a fault message needs to be reviewed by scheduler A (junior) with employee number 1001 and scheduler B (senior) with employee number 1002 before being sent. The engine verifies its digital certificate at each step and signs the message content. The final sent message body contains fault information JSON data and a Base64-encoded digital signature string, generated by scheduler B's private key.

[0092] Therefore, according to the above implementation method, the system can realize intuitive display, interactive review, and multi-channel safe distribution of distribution network grounding fault information through a highly integrated, visualized, and secure human-computer interaction interface, which significantly improves the efficiency and reliability of information flow, while ensuring the traceability and security of the operation process.

[0093] In some embodiments, fault-tolerant control of the processing flow is achieved through a system reliability assurance architecture; the system reliability assurance architecture is configured to: Deploy a periodic polling scheduler to start the data acquisition and fault analysis task pipeline based on a time-triggered mechanism.

[0094] Among them, a periodic polling scheduler refers to a software component that automatically triggers and executes specified tasks according to a preset time interval. A time-triggered mechanism refers to a driving method that initiates task execution based on a system clock or timer. A data acquisition and fault diagnosis task pipeline refers to a processing flow consisting of a series of sequentially executed subtasks, including data acquisition, preprocessing, feature extraction, analysis, and result generation.

[0095] Specifically, the system can be configured with a scheduled task (Job) that runs at a fixed period by introducing a task scheduling framework (such as APScheduler for Python, a lightweight, cross-platform task scheduling library based on the Python programming language). Once the task is triggered, it will sequentially call the data acquisition agent, data preprocessing module, time series pattern recognition algorithm, and multi-source data fusion and analysis algorithm, forming a complete automated processing chain.

[0096] For example, the system is configured to periodically poll the scheduler every 60 seconds. At the exact minute (e.g., 10:00:00), the scheduler automatically starts the task pipeline: first, it collects the latest fault data from the distribution network IV zone system, then performs standardization processing, followed by feature extraction and fault analysis, and finally generates fault information. The entire process is completed within 50 seconds, with the remaining 10 seconds serving as a fault tolerance waiting period.

[0097] An execution timeout detection algorithm is implemented, and an independent execution timeout monitoring channel is established for each processing module.

[0098] Among them, the timeout detection algorithm refers to a mechanism that monitors the execution time of a task or process and intervenes when it exceeds a preset threshold. The execution time limit monitoring channel refers to a monitoring logical path that is configured separately for each critical processing module to track its running status and time consumption.

[0099] Specifically, the system can use multi-threaded or multi-process programming to start an independent monitoring thread or coroutine for each core processing module (such as data acquisition and feature extraction). This monitoring thread starts timing when the module starts and compares the result with a preset maximum allowed execution time (Timeout). If the module does not return a result within the time limit, the monitoring channel will trigger a timeout event. Timeout is a timing control mechanism widely used in computer science, network communication, and software engineering. For example, the system sets a maximum allowed execution time of 5 seconds for the "data preprocessing module" and 10 seconds for the "fault feature extraction module." When each module starts, its corresponding monitoring channel simultaneously starts a timer. If the "data preprocessing module" does not complete its work within 5 seconds, the monitoring channel will mark the module as timed out and notify the subsequent fault-tolerant processing unit.

[0100] Deploy a process status monitor, detect timeout or abnormal exit events by the abnormal status recognition engine module, call the process termination and reconstruction methods to close the associated processes and execute the system re-initialization sequence.

[0101] The process status monitor is a daemon that continuously monitors the running status of critical processes or services within the system. The anomaly detection engine is the logical unit within the monitor that analyzes process status information (such as CPU usage, memory usage, and liveness) and determines whether anomalies have occurred. The process termination and reconstruction method refers to a standard operating procedure that first safely terminates a faulty process and then restarts the process and its dependent components. The system re-initialization sequence refers to a series of initialization steps performed to restore the system to a working state.

[0102] Specifically, the process status monitor can periodically obtain the status information of critical processes through operating system interfaces (such as the `ps` command in Linux) or third-party libraries (such as `psutil` in Python, a cross-platform process and system monitoring library). The abnormal status identification engine determines abnormalities based on preset rules (such as process unresponsiveness, CPU utilization exceeding 95% for 30 seconds). Once an abnormality is detected, the monitor calls a method to forcibly terminate (kill) the process, and then restarts the process and its dependent services or child processes according to the preset startup script sequence.

[0103] For example, the process status monitor detects that the CPU usage of the "feature extraction" process exceeds 150% for five consecutive sampling periods (each period is 2 seconds), which is considered abnormal. The monitor first terminates the process using the kill -9 command, and then executes the following sequentially: (1) Start the data caching service; (2) Start the feature extraction process. The entire restart process is completed within 15 seconds.

[0104] Run an audit trail system to continuously record runtime metrics and abnormal event logs, and establish an operation and maintenance audit data chain.

[0105] The audit trail system is a software subsystem specifically responsible for collecting, storing, and indexing various events during system operation. Runtime metrics refer to performance data generated during system operation, such as task execution time and CPU / memory usage. Anomaly event logs are structured text information recording abnormal states such as system errors, warnings, or timeouts. The operations and maintenance audit data chain links scattered logs and metrics chronologically, forming a continuous data record that can be queried, analyzed, and traced.

[0106] Specifically, the system can embed logging statements at key code points, using standard logging libraries (such as Python's logging module) to write different levels of logs (INFO, WARNING, ERROR) to files or send them to a log server. Simultaneously, it can integrate metric collection libraries (such as Prometheus Client) to record key performance indicators. All this data is timestamped and stored in a database or time-series database, forming a complete audit chain.

[0107] For example, the audit trail system recorded the entire process of a fault assessment task: 10:00:00 Task started (INFO); 10:00:05 Data collection completed, taking 5 seconds (INFO); 10:00:15 Feature extraction module timed out (ERROR); 10:00:18 Process monitor triggered restart (WARNING); 10:00:33 Task successfully completed, total time 33 seconds (INFO). These logs, along with metrics such as average CPU utilization (65%) and memory usage (1.2 gigabytes GB) at the time, are stored in the Elasticsearch database, forming a traceable data chain.

[0108] Therefore, according to the above implementation method, the system can effectively ensure the continuous, stable and reliable operation of the distribution network grounding fault information judgment process through automated periodic scheduling, fine-grained timeout monitoring, proactive process health management and complete operation auditing, and significantly reduce the risk of service interruption caused by software module abnormalities or system crashes.

[0109] In other embodiments, such as Figure 3This diagram illustrates the specific implementation process of the secure data acquisition agent performing real-time fault data acquisition tasks in the above embodiments. Using standardized process symbols, the flowchart clearly outlines the automated, fault-tolerant acquisition process from system login to data parsing, specifically embodying the technical chain of "establishing secure communication → simulating manual operation → performing integrity verification → triggering exception handling." Specifically, the process begins with the step of "launching the browser and logging into the IV zone system," which corresponds to "establishing a secure communication tunnel with the distribution network IV zone system and obtaining a data access token through a credential authentication mechanism." This involves using an automated script to drive a browser instance and injecting an encrypted account and password to complete system authentication and establish a secure session connection. Subsequently, the steps of "navigating to the target data page" and "clicking the export / download button" work together, corresponding to "using a human-computer interaction simulation engine to parse and execute a preset graphical interface operation script, driving the acquisition process based on a predefined rule engine." This means precisely locating webpage elements and triggering data export operations by simulating mouse clicks, keyboard input, and other event sequences, thereby replacing traditional manual clicking and waiting. The crucial "waiting for file generation and saving" and subsequent "success?" checks fully implement the reliability design of "performing hash verification on the collected raw data stream and constructing an end-to-end data security pipeline using a transport layer security protocol." While waiting for the download to complete, the system calculates the file's hash value (e.g., MD5) to verify its integrity and completes transmission via a secure transport layer (TLS) protocol. If verification fails or the process times out, the system enters the "logging error and triggering exception handling" branch, which is directly related to the system reliability assurance architecture in the above embodiment, i.e., ensuring the robustness of the task through exception state identification and handling logic. If successful, the process enters the "parse the downloaded data file" step, which corresponds to "inputting the encrypted real-time fault data into the data preprocessing pipeline," i.e., calling the corresponding data parsing library (e.g., Pandas) to convert the downloaded raw file (e.g., Excel format) into a structured data object for subsequent standardization and filtering. Therefore, Figure 3 The process is a specific and visualized embodiment of the security data acquisition agent technology solution, which clearly demonstrates the closed-loop operation logic from automated login to secure acquisition, and then to integrity protection and anomaly handling.

[0110] Figure 4 This is a structural block diagram of a distribution network grounding fault information analysis system according to an embodiment of the present invention.

[0111] like Figure 4 As shown, the distribution network grounding fault information analysis system includes: The fault dataset construction module 210 is used to acquire multi-source fault data in the distribution network system and preprocess the multi-source fault data to obtain a standardized fault dataset.

[0112] The fault feature information extraction module 220 is used to extract fault feature information from the standardized fault dataset through a time-series pattern recognition algorithm. The time-series pattern recognition algorithm is configured to extract the first half-wave features and measure the waveform similarity of the zero-series current waveform in the standardized fault dataset.

[0113] The fault section location module 230 is used to perform topological clustering analysis on the fault feature information based on the fault feature information through a multi-source data fusion judgment algorithm to generate fault section location results. The multi-source data fusion judgment algorithm is configured to use waveform correlation based on time derivative Euclidean distance to calculate and output section identifiers.

[0114] The grounding fault information generation module 240 is used to generate distribution network grounding fault information based on the topological relationship identifier and electrical parameter characteristics in the fault section location results.

[0115] The specific functions and examples of each module and submodule of the device in this embodiment of the invention can be found in the relevant descriptions of the corresponding steps in the above method embodiments, and will not be repeated here.

[0116] According to embodiments of the present invention, the above-described method of the present invention can be applied to a computer device and a readable storage medium.

[0117] Figure 5 A schematic block diagram of a computer device 600 that can be used to implement embodiments of the present invention is shown. The computer device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The computer device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0118] like Figure 5As shown, the computer device 600 includes a computing unit 601, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. The RAM 603 may also store various programs and data required for the operation of the computer device 600. The computing unit 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0119] Multiple components in computer device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of monitors, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows computer device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0120] The computing unit 601 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as a method for judging distribution network grounding fault information. For example, in some embodiments, a method for judging distribution network grounding fault information can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed on the computer device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the computing unit 601, one or more steps of the method for judging distribution network grounding fault information described above can be performed. Alternatively, in other embodiments, the computing unit 601 may be configured by any other suitable means (e.g., by means of firmware) to perform a distribution network grounding fault information assessment method.

[0121] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0122] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0123] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0124] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0125] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0126] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0127] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0128] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for analyzing distribution network grounding fault information, characterized in that, include: Acquire multi-source fault data from the distribution network system and preprocess the multi-source fault data to obtain a standardized fault dataset; Fault feature information is extracted from the standardized fault dataset using a time-series pattern recognition algorithm, which is configured to extract the first half-wave feature and measure the waveform similarity of the zero-series current waveform in the standardized fault dataset. Based on the fault feature information, a multi-source data fusion and judgment algorithm is used to perform a clustering analysis based on topological relationships on the fault feature information to generate fault segment location results. The multi-source data fusion and judgment algorithm is configured to use waveform correlation based on time derivative Euclidean distance to calculate and output segment identifiers. Distribution network grounding fault information is generated based on the topological relationship identifiers and electrical parameter characteristics in the fault section location results.

2. The method according to claim 1, characterized in that, The standardized fault dataset includes ground fault waveform data; the extraction of fault feature information from the standardized fault dataset using a time-series pattern recognition algorithm includes: The zero-sequence current waveform in the ground fault recording data is subjected to first half-wave truncation processing to obtain the first half-wave waveform; The zero-sequence current fitting function is obtained by performing nonlinear fitting calculation on the first half-wave waveform using a nonlinear least squares optimization algorithm. The time derivative parameters of the fitted waveform are calculated based on the zero-sequence current fitting function, and the Euclidean distance is calculated based on the time derivative parameters to obtain the waveform similarity metric. The fault characteristic information is constructed based on the zero-sequence current fitting function, the time derivative parameter, and the waveform similarity metric.

3. The method according to claim 2, characterized in that, Based on the fault feature information, a multi-source data fusion and judgment algorithm is used to perform topological clustering analysis on the fault feature information to generate fault segment location results, including: Based on the waveform similarity metric, a multidimensional feature vector of the feeder outgoing switch is constructed, and the waveform correlation coefficient between each feeder outgoing switch is calculated using a multidimensional feature distance statistical algorithm. Based on the waveform correlation coefficient, the feeder with the lowest waveform correlation coefficient is identified as the faulty line through feeder-level cluster analysis. Based on the switch topology of the faulty line, the waveform correlation coefficient between adjacent switch nodes in the topology graph is calculated using a segment-level graph analysis algorithm. The algorithm for identifying the least similar node pairs identifies the adjacent switch pairs with the lowest waveform correlation coefficient, generating fault segment location results. The fault segment location results include the identification information of the adjacent switch pairs.

4. The method according to claim 3, characterized in that, The step of generating distribution network grounding fault information based on the topological relationship identifier and electrical parameter characteristics in the fault section location result includes: Based on the identification information of the adjacent switch pairs, a fault section boundary identifier is generated through a boundary node encoding algorithm. The line identifier and fault phase characteristics are obtained from the fault section location results using a feature extraction engine; By combining timestamp information, a standardized fault event description is constructed using an event description generator; The fault section boundary identifier, line identifier, fault phase characteristics, and standardized fault event description are injected into the information fusion engine, and the distribution network grounding fault information is output through standardized encapsulation of the data serialization protocol.

5. The method according to claim 1, characterized in that, The multi-source fault data includes distribution terminal grounding alarm signals, bus grounding information, low-current grounding line selection information, arc suppression coil information, and bus voltage information obtained from the distribution network zone IV system; the multi-source fault data is real-time fault data obtained from the distribution network zone IV system through a safety data acquisition agent, which is configured as follows: Establish a secure communication tunnel with the distribution network IV zone system and obtain data access tokens through a credential authentication mechanism; A human-computer interaction simulation engine is used to parse and execute preset graphical interface operation scripts, and a data collection process is driven by a predefined rule engine. Perform hash verification on the collected raw data stream and construct an end-to-end data security pipeline using transport layer security protocols; The encrypted real-time fault data is input into the data preprocessing pipeline to perform data pattern mapping and transformation, as well as outlier filtering based on statistical distribution.

6. The method according to claim 5, characterized in that, The distribution network grounding fault information is output through a human-machine interface module, which is configured as follows: A graphical monitoring panel is built using a visualization rendering engine to dynamically render the faulty line topology diagram and electrical parameter waveform sequences; It integrates an interactive editing workflow, supports rich text markup language editing, and supports structured data verification protocols; Deploy a multi-channel message distribution engine to encapsulate standardized fault information into a cross-platform message format, adapting to the protocol conversion of SMS gateways and collaborative office systems; Deploy an asynchronous audit workflow engine to integrate multi-level verification nodes into the message distribution pipeline and execute authentication protocols and digital signature algorithms.

7. The method according to claim 1, characterized in that, Fault-tolerant control of the processing flow is achieved through a system reliability assurance architecture; the system reliability assurance architecture is configured to: Deploy a periodic polling scheduler to start the data acquisition and fault analysis task pipeline based on a time-triggered mechanism; Implement a timeout detection algorithm and establish an independent execution time limit monitoring channel for each processing module; Deploy a process status monitor, detect timeout or abnormal exit events by the abnormal status recognition engine module, call the process termination and reconstruction methods to close the associated processes and execute the system re-initialization sequence; Run an audit trail system to continuously record runtime metrics and abnormal event logs, and establish an operation and maintenance audit data chain.

8. A distribution network grounding fault information analysis system, characterized in that, include: The fault dataset construction module is used to acquire multi-source fault data in the distribution network system and preprocess the multi-source fault data to obtain a standardized fault dataset. The fault feature information extraction module is used to extract fault feature information from the standardized fault dataset through a time-series pattern recognition algorithm. The time-series pattern recognition algorithm is configured to extract the first half-wave feature and measure the waveform similarity of the zero-series current waveform in the standardized fault dataset. The fault segment location module is used to perform topological clustering analysis on the fault feature information based on the fault feature information through a multi-source data fusion and judgment algorithm to generate fault segment location results. The multi-source data fusion and judgment algorithm is configured to use waveform correlation based on time derivative Euclidean distance to calculate and output segment identifiers. The grounding fault information generation module is used to generate distribution network grounding fault information based on the topological relationship identifier and electrical parameter characteristics in the fault section location result.

9. A computer device, characterized in that, include: At least one processor; and a memory that is communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, in, Computer instructions are used to cause a computer to perform the method according to any one of claims 1-7.

Images